A lay summary of: Mortality and cancer incidence 1952–2017 in United Kingdom participants in the United Kingdom’s atmospheric nuclear weapon tests and experimental programmes.

Reference: Michael Gillies and Richard G E Haylock (2022) Mortality and cancer incidence 1952–2017 in United Kingdom participants in the United Kingdom's atmospheric nuclear weapon tests and experimental programmes.Journal of Radiological Protection, 42(2), 021528. DOI 10.1088/1361-6498/ac52b4


What were the research questions?

The Nuclear Weapons Test Participant Study is assessing the health impact of being present at sites used in the British nuclear testing programme. The study has previously reported on three occasions. The first presented information on the numbers of death and cancers documented in national registers up in 1983, the second up to 1990 and the third up to 1998. The findings of these studies revealed no detectable effect on overall life expectancy or, on total risk of cancer, when compared to a control group of veterans. The studies did show some evidence for a small excess in leukaemia, particularly in the early years after the tests which the authors describe as meaning a small increased risk of leukaemia could not be ruled out.

The purpose of the work reported by Gillies and Haylock is to strengthen the statistical power and the certainty in understanding by increasing the follow-up period to 2017. This adds an additional 19 years of reported mortality (deaths) and cancer incidence to the epidemiological dataset and with an average veteran age of 81, would represent the lifetime impact of test participation on health.

How was the scientific problem approached?

An epidemiological case-control study was set up and previously described by colleagues of the authors. This included military veterans from all services and civilian employees of Atomic Weapons Establishment (AWE) and Atomic Energy Research Establishment who had been present at one or more nuclear test sites or, had sampled radioactive plumes. A total of 21,357 test participants were verified mainly through information obtained from MoD records. A control group of 22,312 veteran or AWE personnel who did not participate in the tests, who were matched for age, service, rank and date of entry to the study, were also recruited to the epidemiological study.

What did the research involve?

Information on the number of deaths, including their cause of death and, the number diagnosed with cancer was determined for all individuals within the test and control group. This information was obtained from national registries.
These numbers were then compared in two different ways.

  • Numbers in test and control groups were compared with expected numbers of deaths and cancers assuming similar rates to the general population. These are known as standardised mortality ratio (SMR) or standardised incidence ratio (SIR).
  • Numbers were compared directly between test and control groups. This is expressed as a relative risk (RR) or risk in test participants relative to control group.

For each of these statistical comparisons, some sub-group analysis to examine for specific cause of death or cancer sub-types were performed in addition to sub-group analysis considering recorded radiation exposure levels and/or different high-risk job roles.

What did they find?

Compared to general population:
The proportion of deaths in the test participants is lower than that expected in the general population (SMR=90, p<0.001). This was due to their being less non-cancer related deaths (SMR=85, p<0.001). The number of cancer related deaths (SMR=97, p=0.08) and cancer diagnosis (SIR=101, p=0.27) was similar to those seen in the general population.

Compared to control veterans and AWE workers
A small increase (2%) in number of deaths in test participants was recorded compared to controls (RR=1.02, 90% CI 1.00-1.05, p=0.04). The number of deaths from cancer was 3% higher (RR 1.03, 90% CI 1.00-1.07) while for non-cancer diseases the figure was 2% (RR=1.02, 90% CI 1.00-1.05).

Several specific cancers contributed to these higher rates including non-CLL leukaemia, cancers of the stomach, bladder, prostate and respiratory organs.

The diagnosis of chronic myeloid leukaemia (CML) is higher in the test participants in comparison with both the national population (SIR=151, p=0.03) and the control group (RR 2.43, 90% CI 1.43-4.13, p=0.003).

Prostate, brain and non-melanoma skin cancer also showed some evidence of being reported more in sub-groups who had a greater potential for exposure.

For non-cancer diseases, the increase in number of deaths was mainly due to cerebrovascular disease.

How did the researchers interpret their results?

The authors highlight that differences in characteristics, such as smoking habits, between the participation and control groups could be an explanation for the small increases in death and disease.

The authors suggest their findings indicate the potential for test participation to lead to a small increase in risk of non-CLL leukaemia cannot be ruled out. They highlight the risk for this appears stronger for CML whilst noting this is a rare disease with only 32 instances observed among the participants and 14 among the controls.

The authors state there was some evidence that the number of reported incidences is increased in those participants most likely to have been exposed to radiation and, in those groups most likely to have been exposed to internal emitters. However, they caution interpretation as the actual numbers of individuals in these sub-groups is very low.

Overall, they state that a small impact of test participation for the small increases in death and disease cannot be ruled out.

Who did this research?

Researchers of the UK Health Security Agency carried out this work.

Please let us know what you think. Did you find this useful?

Click on a star to rate it!

We are sorry that this page was not useful for you!

Tell us how we can improve this?

Supporting Resources

Key Messages

The study examines the lifelong impact of participation at test sites

Number of deaths registered among the test participants (and the controls) is lower than expected from national rates

Number of deaths registered in participants is slightly (2%) higher than in the control group

Evidence for an increased risk of leukaemia, particularly CML, in comparison with both the national population and the control group

Some evidence that participants had higher mortality rates from cerebrovascular diseases

Background characteristics of the participant and the control group (e.g. smoking habits, diet) are not known and could be responsible for the differences reported

A small long-term detrimental health effect associated with participation in the tests cannot be completely ruled out


Links to the research paper

This is a peer-reviewed study meaning that other scientists have reviewed this work before the authors published it in the Journal of Radiological Protection in 2022.